Climate change and other forms of hydrologic non-stationarity are creating new challenges for the science and effective practice of environmental flows. Foundational assumptions underlying standard e-flows approaches are based in the notion of a reference state, both hydrologically and ecologically. Shifting statistical baselines in hydro-ecosystems due to increasing water demand and climate change require a re-thinking of the methods and measures of success in e-flows. The prevailing paradigm of a restoration focus based on statistical averages of long-term hydrologic records and presumed equilibrial ecosystem states is giving way to a forward-looking adaptation approach, where hydrosystems are viewed as event-driven and ecologically dynamic. New concepts about managing for resilience or against failure are emerging to grapple with climatic non-stationarity. Emerging imperatives for e-flows are broadening of targets (to include states, rates and traits), embracing uncertainty in endpoints, and adopting a more process-based (as opposed to statistical) approach that utilizes monitoring to gauge success in terms of system trajectories in an adaptive fashion. Given the limited water typically available for e-flows, prioritization of e-flows interventions is increasingly needed and careful targeting of e-flows is necessary to support economically efficient yet ecologically effective management. Beyond these concerns, the traditional e-flows focus on at-a-site restoration needs to be expanded to a broader whole-basin water management perspective to address the scales of human alteration of river systems relevant to long-term freshwater conservation.